INTRODUCTION

Tumor necrosis factor  (TNF)¹ is a cytokine produced by activated macrophages as well as by several other cell types, including lymphocytes, fibroblasts, and hepatocytes (reviewed in Refs. 1-3). The effects of TNF are mediated by two distinct cell surface receptors, p55 TNF-R1 and p75 TNF-R2, that are expressed on almost all nucleated cells (3). Both receptors lack recognizable enzymatic domains, and their ability to transduce signals is dependent on the interaction with proteins associating with their cytoplasmic tails (4, 5). TRADD (an acronym for TNF receptor 1-associated death domain-containing protein) has been identified as a TNF receptor 1-associated protein that contacts the TNF-R1 death domain in a ligand-dependent manner (6). TRADD acts as an adapter (7) whose function is to recruit to the receptor the death domain protein FADD (8, 9) and TRAF2 (10). Although FADD interacts with and activates the apoptotic proteases (11, 12), TRAF2 is required for the activation of both transcription factor NFB (7, 13) and the c-Jun N-terminal kinases (also known as stress-activated protein kinases; JNK/SAPK) (14-16), a family of proline-directed Ser/Thr kinases (17-19) that bind, phosphorylate, and activate the transcriptional activation domains of c-Jun, ATF2, and Elk1 (20-22). Therefore, although TNF-R1-induced apoptosis requires FADD, stimulation of gene expression mainly depends on TRAF2. p75 TNF-R2 does not recruit TRADD but directly interacts with TRAF2 (10); therefore it is able to activate NFB and JNK/SAPK in a TRAF2-dependent manner (13, 16), but it does not induce FADD-dependent apoptosis.

Signaling events downstream of TRAF2 are largely unknown. The protein kinase NIK has been recently identified and cloned as a TRAF2-interacting protein that is both sufficient and required for NFB activation (23). However, the role of NIK in additional TNF-R1 pathways is not known. We therefore investigated the signaling properties of NIK and its role in JNK/SAPK and p38 activation by TNF-R1.

EXPERIMENTAL PROCEDURES

Full-length human NIK cDNA was PCR amplified from a human placental cDNA library using a mixture of Taq and Pwo polymerases (Boehringer). The primers used were NIK1 (5-TCCGCTAGCATGGCAGTGATGGAAATG-3) and NIK2 (5-CTCTCTAGATTAGGGCCTGTTCTC-3); the PCR fragment was digested with NheI and XbaI and cloned in pCDNA3-HA. pCDNA3-HA was constructed by insertion of a BglII/BamHI fragment from pActII (CLONTECH) into the BamHI site of pCDNA3 (Invitrogen, Inc). NIK1234 (deletion of aa 1-334) was constructed by PCR amplification of full-length NIK using the primers 1234 (5-TTGCTAGCGAATACCTAGTGCATGCTCTG-3) and NIK2. NIK2101 (deletion of aa 1-623) was amplified using the primers 2101 (5-TTGCTAGCATGCCTCTCACAGCCCA-3) and NIK2. Both PCR fragments were cloned in pCDNA3-HA as above. NIK-KR was obtained by two-step PCR using the mutagenic primers KRS (5-CAGTGCGCTGTCAGGAAGGTGCGGCTGGA-3) and KRAS (5-CAGCCGCACCTTCCTGACAGCGCACTG-3). HA-p46SAPKpCDNA3 and HA-SEKALpMT2 (gifts of J. R. Woodgett), FL-mTRAF2pRK and FL-mTRADDpRK (gifts of David V. Goeddel) have been already described (6, 13, 15, 32). The reporter plasmids NFB-CAT and -73ColCAT (gifts of Michael Karin) have been described in Refs. 24 and 25. p38(FLAG)pCDNA3 is a kind gift of R. J. Ulevitch.

293 cells were transfected with the calcium phosphate coprecipitation method using 8 µg of total plasmid DNA, whereas HeLa cells were transfected with LipofectAMINE (Life Technologies, Inc.). 48 h after transfection, cells were lysed in radioimmune precipitation buffer containing 0.5 mM DTT, 20 mM -glycerophosphate, 1 mM sodium orthovanadate, 10 mM sodium fluoride, 1 mM phenylmethylsulfonilfluoride, 20 µg/ml leupeptin, 20 µg/ml aprotinin.

Lysates were cleared by centrifugation, and protein concentration was measured using a commercial Bradford protein assay (Promega). Equal amounts of each lysate (usually 500 µg) were incubated on ice with 2 µg of anti-HA antibody 12CA5 (Boehringer) (JNK assays) or anti-FLAG antibody (IBI Kodak) (p38 assays) for 2 h. Immune complexes were collected by protein A-agarose for 25 min, washed thrice with radioimmune precipitation buffer containing 20 mM -glycerophosphate, 1 mM sodium orthovanadate, 0.5 mM DTT, and washed once with kinase reaction buffer (20 mM Hepes, pH 7.5, 20 mM MgCl₂, 20 mM -glycerophosphate, 2 mM DTT, 100 µM sodium orthovanadate, 0.5 mM sodium fluoride). Samples were finally resuspended with 40 µl of kinase reaction buffer containing 20 µM ATP, 2.5 µCi of [-³²P]ATP and either 2 µg of glutathione S-transferase-c-Jun (1-141) (JNK assays) or 8 µg of mielin basic protein (p38 assays) and incubated at 30 °C for 20 min. Reactions were stopped by the addition of 3 × Laemmli sample buffer; samples were boiled and loaded on 12.5% SDS-acrylamide gels. After fixing and drying, gels were autoradiographed at 70 °C. Radioactivity in each spot was quantitated with a PhosphorImager. The amount of exogenous transfected kinase in each sample was analyzed by Western blotting. Reporter gene assays were performed as described (24).

RESULTS AND DISCUSSION

In both 293 and HeLa cells, TRADD, TRAF2, and NIK are able to induce a strong NFB activation (NIK being the strongest activator) (Fig. 1, A and B). Deletion of the N-terminal (putatively regulatory) domain of NIK (NIK1234) does not apparently affect NFB activation, whereas the deletion of the catalytic domain (NIK2101) or its inactivation (Lys  Arg mutation at aa 429) abolish NFB activation (Fig. 1B). Consistent with a requirement for NIK in TNF-induced NFB activation, overexpression of a C-terminal NIK fragment (NIK2101), which binds TRAF2 and presumably blocks the recruitment of endogenous NIK and/or titrates downstream effectors (23), significantly impairs the induction of NFB by either TNF treatment or overexpression of TNF-R1 complex components (Fig. 1C). These data indicate that NIK is required for the activation of NFB by TNF-R1/TRAF2 in different cell types.

NIK mediates activation of transcription factor NFB by TNF receptor 1. 

NIK1234

NIK2101

Because TRAF2 overexpression is sufficient to activate both NFB and JNK/SAPK (13-16), we examined whether NIK is able to activate JNK/SAPK as well. NIK was cotransfected in HeLa and 293 cells together with a hemagglutinin (HA)-tagged SAPK expression vector, and the activity of exogenous transfected SAPK was assayed 36-48 h after transfection. In 293 cells, wherein TRAF2 usually gives the highest activation of SAPK/JNK, neither NIK or NIK1234 (which are both efficient NFB activators) were able to elevate JNK/SAPK activity over the baseline. In a similar manner, we were unable to detect any JNK/SAPK activation by NIK in HeLa cells (Fig. 2, A and B). Similarly to JNK/SAPK, p38 activation by TNF depends on TRAF2 (14)² but is not dependent on NIK (Fig. 2C).

Apart from inducing a prolonged activation of JNK/SAPK and p38, TNF-R1 engagement provokes a mild and transient activation of the mitogen-activated protein kinase (MAPK), whose biological role has not been defined (26, 27). MAPK activation by TNF may depend on a TNF-R1 domain that is distinct from the TRADD interaction domain and that interacts with a recently identified protein known as FAN (28). Consistent with MAPK activation being a TRADD-independent function, neither TRAF2 or NIK were able to activate MAPK in the cells tested (Fig. 2D).

The effects of the dominant negative NIK mutant (NIK2101) on JNK/SAPK activation by TNF-R1 were next evaluated. The expression of 2101 at levels that gave maximal inhibition of NFB induction (Fig. 1C) did not impair the ability of either TNF or TRAF2 to activate JNK/SAPK (Fig. 2E). Therefore, when the NIK pathway is blocked by expression of dominant negative NIK, both receptor cross-linking and overexpressed TRAF2 still activate JNK/SAPK. Taken together our results suggest that: (i) NIK is neither sufficient nor required for JNK/SAPK activation by TNF-R1/TRAF2; (ii) the bifurcation between the NFB pathway and the JNK/SAPK pathway occurs immediately downstream of TRAF2; and (iii) dominant negative NIK does not disrupt the TNF-R1 complex nonspecifically. Therefore, NIK dissects the TRAF2 pathway leading to NFB activation from the pathway leading to JNK/SAPK activation; this suggests that the ability of TNF-R1/TRAF2 to activate JNK/SAPK must depend on a different TRAF2-interacting protein. One possible candidate is represented by MEKK1, a kinase that phosphorylates and activates SEK/JNKK, which in turn phosphorylates JNK/SAPK (29-32). However, we have been unable to detect a physical interaction between TRAF2 and MEKK1.² Therefore, the evidence for a role of MEKK1 in TNF-R1 signaling is indirect and arises from the ability of catalytically inactive MEKK1 (MEKK1-KM) to block TNF-R1/TRAF2-induced activation of SAPK/JNK (14); at this point we cannot exclude the possibility that SAPK/JNK activation by TNF-R1/TRAF2 depends on a putative MEKK1-related protein whose activity is inhibited by MEKK1-KM overexpression.

The prolonged activation of JNK/SAPK by TNF and the consequent phosphorylation and activation of the c-Jun transcriptional activation domain correlate with the sustained induction of AP1-dependent genes (33, 34). AP1 is composed of proteins of the Jun and Fos families that associate to form a variety of homo- and heterodimers that bind to a common recognition element known as either the tetradecanoic phorbol acetate-response element or the AP1 binding site (35); the presence of AP1 sites in the promoters of several genes, including those encoding for cytokines and adhesion molecules, contributes to the induction of such genes by TNF as well as by other AP1 inducers. With respect to TNF, TRAF2 (which is an efficient JNK/SAPK activator) was found to be a stronger stimulator of AP1-dependent transcription (Fig. 3A). Unexpectedly, overexpression of NIK, which activates neither JNK/SAPK nor MAPK, strongly activated transcription directed by a canonical AP1 site. This effect of NIK was dependent on an intact protein kinase domain. Moreover, AP1 activation by NIK was not blocked by dominant negative SEK/JNKK (Fig. 3B) or by chemical inhibitors of p38 and MAPK/Erk.² Evidence that NIK contributes to the induction of AP1 activity by TNF-R1/TRAF2 comes from experiments with dominant negative NIK; indeed, NIK2101 was able to reduce TNF-induced activation of AP1 by more than 50% without any evident effect on basal AP1 activity (Fig. 3C). Therefore, when the NIK pathway is blocked and the JNK/SAPK pathway fully active (Fig. 2), the ability of TNF-R1 to stimulate AP1 activity is severely impaired. The effect of NIK2101 on TRAF2-dependent activation of AP1 was slightly less evident; this may reflect a major contribution of the JNK/SAPK pathway to AP1 activation by overexpressed TRAF2, consistent with the greater potency of transfected TRAF2 compared with TNF in JNK/SAPK induction (Fig. 2).

TRAF2 is a critical signaling molecule that links both p55 TNF-R1 and p75 TNF-R2 to NFB and JNK/SAPK activation pathways. TRAF2-dependent activation of NFB is required for the induction of several genes, including those protecting cells from TNF-induced apoptosis (14, 15, 37-39); conversely, the activation of JNK/SAPK does not seem to be relevant for cytotoxicity (14, 15) but collaborates to the induction of adhesion molecules and cytokines (40).² The results reported here indicate that the NFB and the JNK/SAPK pathways bifurcate immediately downstream of TRAF2; this would suggest that TRAF2 signals by interacting with and activating at least two distinct effectors, namely NIK, which seems to be responsible for NFB activation (possibly through direct activation of the recently identified IB kinase) (41, 42), and yet unknown transducers are responsible for JNK/SAPK activation. Apart from inducing NFB, NIK couples TNF-R1/TRAF2 to AP1 activating pathways that are alternative to MAPK and JNK/SAPK. The inhibitory effect of dominant negative NIK on AP1 activation by TNF suggests that this AP1 activation pathway is not redundant but probably collaborates with the JNK/SAPK and the MAPK pathways to achieve an optimal AP1 activation. The mechanisms of AP1 activation by NIK is still unclear. Because AP1 is a collection of dimers composed by several Fos and Jun family proteins, it is highly likely that a number of regulatory mechanisms other than c-Jun phosphorylation contribute to control its activity at various levels. The identification of downstream target(s) of NIK will help elucidate the mechanism of AP1 activation through this pathway.